AI Safety Seeding Initiative launches to build university groups

The AI Safety Seeding Initiative, announced in partnership with Kairos, seeks to expand the number of AI safety university groups by identifying and supporting student founders at schools that currently lack them. The initiative's core insight is grounded in empirical evidence: a 2023 survey of people working on catastrophic risks found that university group involvement was the single most common career influence in the field, demonstrating the outsized importance of these grassroots communities in producing AI safety talent.

Despite their proven value, the initiative notes that dozens of top universities without existing AI safety groups could sustain them—a gap in the talent pipeline the program aims to address. The initiative will identify latent founders at these schools, support them through the group's earliest stages, and prepare them to apply to Pathfinder (suggesting a competitive fellowships or grants program) with a goal of launching this fall. The program is also calling on the community to help by referring potential student founders at universities without AI safety groups; a list of eligible schools is available on the initiative's website at aisafetyseeding.org.

Existing groups supported by Kairos, such as MAIA, AISST, and BASIS, have grown from small beginnings to become established communities within their universities, providing a model for what the initiative hopes to replicate. By formalizing the process of identifying and supporting new founders, the AI Safety Seeding Initiative aims to accelerate the creation of groups at institutions where the groundwork has not yet been laid.

University-based AI safety groups have emerged as a critical talent pipeline for the field. According to a 2023 survey cited in the announcement, group involvement was the single most common career influence among people working on catastrophic risks—a finding that underscores the outsized impact these grassroots communities have in shaping career trajectories. Yet despite their proven value, dozens of universities with the resources and academic strength to sustain such groups remain without one, representing a significant untapped opportunity for field building.

The AI Safety Seeding Initiative addresses this gap by targeting the earliest bottleneck: identifying and supporting potential founders before they launch. By partnering with Kairos, an established field-building organization that already supports existing groups like MAIA, AISST, and BASIS, the initiative can leverage existing expertise and networks to accelerate the creation of new groups. The focus on preparing founders for a Pathfinder application and a coordinated fall launch suggests a structured approach designed to increase the likelihood of sustainable group formation.

No comments yet. Be the first to share your thoughts!